Review and update overhead model

[dkirkby]

From a recent email by DJS to desi-data:

The Mayall with the upgraded TCS has overheads that are well-fit with
   Overhead = 24.5 + 2.3 * Slew [sec, for slews < 10 deg]
                    = 27 + 2.3 * Slew [sec, for slews > 10 deg]
where Slew is the maximum of the HA and Declination slews.
I don’t know the reason for the discontinuity at 10 deg slews.

This looks to be ~17 sec more than what’s currently being assumed.

The scatter of some overhead times to larger values may be due to
dome slews; I’ve not dug into those cases.

Dome slews are another item that we should be considering.  We ignored
this for the MzLS imaging survey, yet if one’s in a situation of observing
multiple fields right around HA=0, one could find oneself rotating the dome
180 deg between exposures.  It may be enough to have the afternoon
planning optimize some of these out.

[dkirkby]

Details of the overhead model based on MzLS experience are now in DESI-3687.

[dkirkby]

The scheduler currently assumes a fixed 120s overhead between exposures of different fields, based on my understanding that slew time is unlikely to dominate. This might change, so I am leaving this issue open for now, as a reminder to revisit this once the ICS inter-exposure timing is finalized.